Analysis of dose-response effects on gene expression data with comparison of two microarray platforms

Area of Science:

• Biomedical research
• Genomics
• Statistical bioinformatics

Background:

• Analyzing dose effects on gene expression is critical in biomedical research.
• Identifying genes with non-random, dose-dependent expression changes is a key challenge for biomarker discovery.
• Existing methods may not fully capture monotonic dose-response relationships.

Purpose of the Study:

• To formally apply and evaluate an isotonic (monotonic) regression approach for analyzing gene expression dose-response.
• To introduce a novel test statistic for selecting genes with significant monotonic dose-response expression.
• To compare the performance of isotonic regression with a likelihood ratio-based test.

Main Methods:

• Application of isotonic regression for dose-response analysis in gene expression studies.
• Development of a permutation-based test statistic to identify monotonic gene expression trends.
• Comparative analysis against a traditional likelihood ratio-based test.

Main Results:

• The isotonic regression method was successfully applied to gene expression data from RKO colon carcinoma cells treated with 5-fluorouracil.
• The study allowed for a comparison of gene expression data from Affymetrix and printed 75mer oligomer cDNA arrays.
• The effectiveness of the isotonic regression approach in detecting monotonic dose-response patterns was demonstrated.

Conclusions:

• Isotonic regression provides a robust statistical framework for analyzing gene expression dose-response relationships.
• This approach aids in the identification of potential biomarkers exhibiting monotonic expression changes.
• The developed statistical software is available for implementation in similar biomedical studies.